PICC

A collection of macOS automation & productivity tools written in Rust, built on Apple native frameworks (CoreGraphics, Vision, AppKit, Accessibility) via the objc2 ecosystem.

Tools

picc — Screenshot + OCR

Interactive screenshot tool with region selection and text recognition.

• Ctrl+Cmd+A global hotkey to activate
• Drag to select region, real-time pixel resolution display
• OCR powered by Vision framework (Chinese + English)
• Runs in background with no Dock icon

cargo run --bin picc

axcli — Accessibility CLI

Playwright-style CLI for macOS app automation via the Accessibility API. Supports element locating, clicking, typing, scrolling, screenshots with OCR, and more.

axcli --app Lark snapshot              # Print accessibility tree
axcli --app Lark click '.SearchButton' # Click element by locator
axcli --app Lark input '.SearchInput' 'hello'
axcli --app Lark screenshot --ocr -o /tmp/shot.png
axcli --app Lark get AXValue '.SearchInput'

standup — Break Reminder

Pomodoro-style break reminder with menubar countdown. Supports LAN sync across multiple devices via UDP broadcast.

cargo run --bin standup -- --work 25 --break 5
cargo run --bin standup -- --solo  # Disable LAN sync

ax_print / ax_tui — Accessibility Tree Viewers

Print or interactively browse the accessibility tree of any macOS app.

cargo run --bin ax_print -- --app Lark
cargo run --bin ax_tui -- --app Chrome

hidden — Menubar Manager

Hidden Bar clone — collapse/expand menubar icons by toggling a separator status item.

cargo run --bin hidden

dictation — Push-to-Talk Dictation

Hold right Command key to dictate speech, recognized text is typed at the cursor. Supports SenseVoice (offline, via sherpa-onnx) and Apple Speech API engines.

# Install
cargo install --git https://github.com/andelf/picc --bin dictation --features sensevoice

# Apple Speech API (default, no setup needed)
dictation --engine apple

# SenseVoice (offline, model auto-downloaded on first run ~250MB)
dictation --engine sensevoice

dictation-ng — Push-to-Talk Dictation (Fun-ASR-Nano) [Not Recommended]

Alternative dictation tool using Fun-ASR-Nano (0.8B LLM-based ASR, 31 languages). Inference is significantly slower than SenseVoice due to the larger LLM-based model. Use dictation --engine sensevoice instead.

# Build (requires sherpa-onnx v1.12.28 static library at ~/.local/share/picc/sherpa-onnx-v1.12.28/)
cargo run -p dictation-ng

# Model auto-downloaded on first run (~715MB)

voice-correct — Voice Dictation + Correction

Hold right Command to dictate speech; tap then hold to voice-correct existing text using LLM. Combines push-to-talk dictation with voice-driven text correction powered by Kimi API.

• Hold right Cmd: dictation mode — speak and text is typed at cursor (zero-latency recording)
• Tap + hold right Cmd: correction mode — speak a correction instruction (e.g. "put a period at the end"), LLM modifies the focused text field
• Audio feedback (system sounds) for start/stop/error
• Animated menubar status icon: idle / recording / correcting / processing
• Smart text replacement: AX API for native apps, clipboard paste for browsers/Electron/Lark
• Terminal-aware: skips AX text reading for Terminal, iTerm2, Alacritty, Ghostty, etc.
• Supports both Apple Speech API and SenseVoice (offline) engines

# Apple Speech (default)
KIMI_API_KEY=sk-... cargo run --bin voice-correct

# SenseVoice (offline)
KIMI_API_KEY=sk-... cargo run --bin voice-correct --features sensevoice -- --engine sensevoice

claude_menubar — Claude Code Status Hook

Menubar indicator showing Claude Code session status, driven by Claude Code hooks.

cargo run --bin claude_menubar

Install

# All tools (without dictation, fast build)
cargo install --git https://github.com/andelf/picc

# All tools including dictation (slower, compiles sherpa-onnx)
cargo install --git https://github.com/andelf/picc --features sensevoice

Requirements

• macOS
• Accessibility permission (System Settings > Privacy & Security > Accessibility)

License

MIT/Apache-2.0